Habitual physical activity (PA) provides numerous health benefits for children and adolescents. Average PA levels substantially decrease during adolescence, which influences long-term health conditions and healthcare costs. A few studies have suggested that PA behavior may be established in different times and in different patterns by child subgroups. Studies have also examined the determinant of PA behavior change. However, the analytic approach of the most studies has not appropriately captured the complex relationship among multiple bio-social factors hypothesized to affect PA behavior. The knowledge gap and methodological challenge can be addressed using an innovative analytic approach such as latent class growth analysis and latent variable analysis the objective of this project is to understand the development of PA behavior throughout childhood and bio- social influences of the PA development. We hypothesize that the development of child PA behavior is characterized by multiple developmental pathways representing distinct groups within the child population. We also hypothesize that there are underlying bio-social characteristics that account for PA change patterns during childhood and early adulthood. To test these hypotheses, we propose to conduct secondary data analysis using the longitudinal database from the Iowa Bone Development Study database. The three specific aims of this project are AMI 1) to determine developmental trajectories of MVPA behavior from age 5 to 19 years, AIM 2) to identify bio-social characteristics associated with different developmental trajectories of MVPA behavior identified in AIM 1, and AIM 3) to determine the trajectories for overall PA and sedentary time, and identify their associated bio-social characteristics. We will use latent class growth analysis to identify multiple developmental trajectories of PA. We will also use a latent variable approach to characterize the subpopulations following a certain PA trajectories. This study will expand knowledge on patterns and determinants of PA behavior. The success of this project will accelerate the use of the innovative latent class methodology in pediatric PA research.